04/12/2025
How can AI-generated images better reflect the diverse realities of our world?....
'VisionRAG', developed by Judd Smith @ and Virginia Zangs , tackles biases in AI vision models by integrating Google Street View with LLaVA and SDXL-Turbo.
Their tool generates more culturally accurate images, challenging the Westernized defaults of AI datasets. In an era where synthetic media shapes perception, their work is a vital step toward more inclusive and trustworthy AI representations in journalism, design, and beyond.
✨Created during the Hacks/Hackers & Brown Institute for Media Innovation Hackathon at Columbia University, 2024.
This work will be exhibited as part of the Venice Biennale Architettura 2025
📍 Explore the full project in our archive—link in bio!